Merge pull request #1 from cskiraly/arm64

add support for Apple M1 and other arm64+Neon architectures

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "sse2neon"]
+	path = sse2neon
+	url = https://github.com/DLTcollab/sse2neon

LeopardCommon.cpp

@@ -107,6 +107,8 @@ static void _cpuid(unsigned int cpu_info[4U], const unsigned int cpu_info_type)
 #endif
 }
 
+#elif defined(LEO_USE_SSE2NEON)
+bool CpuHasSSSE3 = true;
 #endif // defined(LEO_TARGET_MOBILE)
 
 

LeopardCommon.h

@@ -186,6 +186,11 @@
 // Unroll inner loops 4 times
 #define LEO_USE_VECTOR4_OPT
 
+// MacOS M1
+#if defined(__aarch64__)
+  #define LEO_USE_SSE2NEON
+  #define LEO_TARGET_MOBILE
+#endif
 
 //------------------------------------------------------------------------------
 // Debug
@@ -256,6 +261,8 @@
     // Note: MSVC currently only supports SSSE3 but not AVX2
     #include <tmmintrin.h> // SSSE3: _mm_shuffle_epi8
     #include <emmintrin.h> // SSE2
+#elif defined(LEO_USE_SSE2NEON)
+    #include "sse2neon/sse2neon.h"
 #endif // LEO_TARGET_MOBILE
 
 #if defined(HAVE_ARM_NEON_H)
@@ -270,6 +277,8 @@
     // Compiler-specific 128-bit SIMD register keyword
     #define LEO_M128 uint8x16_t
     #define LEO_TRY_NEON
+#elif defined(LEO_USE_SSE2NEON)
+    #define LEO_M128 __m128i
 #else
     #define LEO_M128 uint64_t
 # endif
@@ -335,6 +344,8 @@ void InitializeCPUArch();
 # endif
     // Does CPU support SSSE3?
     extern bool CpuHasSSSE3;
+#elif defined(LEO_USE_SSE2NEON)
+    extern bool CpuHasSSSE3;
 #endif // LEO_TARGET_MOBILE
 
 

sse2neon

@@ -0,0 +1 @@
+Subproject commit cad518a93b326f0f644b7972d488d04eaa2b0475